add ability to decorate tracks with "non-genic" content

[adf-ncgr]

I think we've discussed this in the past, but I didn't see an issue commemorating it. The basic idea is that there are things like repeats (or other features that may be of interest in some specialized contexts such as transcription factor binding motifs) which probably shouldn't participate in defining the similarity/alignment of tracks but whose presence/absence between tracks could be of interest. I could imagine treating orphan genes as being in this category, since they can't really contribute to alignment matching. Note the spiffy new gcv3 label I've added for this @alancleary !

[alancleary]

Thanks. I've added this to the version 3 bucket.